The 1st AI-native bioreactor

The full power of AI for the biotech industry.

We build next-generation bioreactors designed to unlock the full potential of AI. For you it stays simple: set your goal, start the run, and the AI keeps it on track.

  • Plugs AI into your existing workflow
  • Saves months of research
  • Moves you faster from lab to market
The product

One bioreactor. AI in every layer.

We design the hardware and the AI together, so every model gets the data it needs.

Lab-scale bioreactor

Built for data: every sensor placed where the AI needs it.

Soft sensors

Turns raw signals into live concentrations, without taking samples.

Digital twin

Gives the AI live digital context during every experiment.

AI process control

Adjusts feed, oxygen, pH and temperature live, before problems start.

Experiment planning

Picks the next most informative run, so you reach answers with fewer experiments.

AI co-pilot

Ask your data. It explains runs, spots patterns and drafts reports.

Reporting & planning

Turns run data into techno-economic analysis (TEA), life-cycle assessment (LCA) and clear scale-up decisions.

How the layers work together in one run

Illustrative example
  1. 1MeasureBioreactorSoft sensors
    pHO₂TempCO₂Feedand many moreLive data every secondSoft sensorCell growth4.2 g/LProduct3.1 g/Llive · every 5 min · no sampling

    Every second the bioreactor logs a rich stream of data. Soft sensors turn it into live cell growth and product levels, with no sampling.

  2. 2PredictDigital twin
    O₂critical < 30%0 h12 h24 hnow · 14 hbelow 30% at 18 hmeasuredtwin forecast

    The digital twin gives the AI live context on the culture. At hour 14 it shows oxygen heading below 30% by hour 18.

  3. 3ActAI process control
    1 · AlertTwin alertOxygen will dropbelow 30%in 3 hoursat 14 h2 · AI adjusts · 15 hAirflow↑ +40%Feed↓ −20%○ before ● afterfully automatic3 · Result · 20 hWithout AIO₂ 22%✕With AIO₂ 40%✓Growth continues

    The AI acts before the problem appears. It raises airflow and lowers feed on its own, so oxygen stays at a safe level.

  4. 4Learn & reportExperiment planningAI co-pilotReporting
    40 possible next runs3 chosen: most to learnfor next weekWhy did yield rise vs run 14?The oxygen limit was avoidedat hour 15.run 14run 15yieldPlan runsReport ready ✓

    After harvest, the planner picks the 3 most useful runs out of 40. The co-pilot explains the result and drafts the run report for your team.

What it unlocks

The technologies of the future, ready today.

Built for the boldest ideas, the ones that are hardest to control.

Synthetic biology

Challenge

New engineered strains behave unpredictably.

With Cellium

Data-rich runs shorten every design–build–test–learn cycle.

Precision fermentation

Challenge

Proteins and ingredients need high titres at low cost.

With Cellium

AI control keeps every run in its most productive window.

Microbial consortia

Challenge

Several species in one vessel drift out of balance.

With Cellium

The twin tracks the community and steers it back.

Pharma & biologics

Challenge

Strict quality demands identical, documented runs.

With Cellium

Closed-loop control and full data records make runs reproducible.

Our mission

Boosting the bioeconomy together.

Bio-based products can replace fossil chemicals, materials and food ingredients. But developing a bioprocess is still slow, costly and risky. We remove the bottlenecks.

What holds bioprocesses back

  1. Slow cycles

    A run takes days and often answers one question.

    CelliumAI picks the runs that teach the most.

  2. Data-poor runs

    A few signals and manual samples per run.

    CelliumRich, live data from every run.

  3. Unpredictable biology

    Living cells drift. One bad hour can ruin a batch.

    CelliumThe twin gives the AI context to act early.

  4. Risky scale-up

    What works in the lab often fails in production.

    CelliumLab data carries the knowledge forward.

  5. Late economics

    Cost and climate impact are known too late.

    CelliumTEA and LCA from the first runs.

  6. Expert-dependent

    Processes rely on a few specialists watching the vessel.

    CelliumSimple to run, with a co-pilot that shares the know-how.

It starts in the lab

Limited data hides errors and gaps in the lab, and they are carried into scale-up. Cellium closes them before you scale.

time &moneyspent Lab research Scale-up Errors & gaps pile up With Cellium Today Ready to scale Errors & gaps Illustrative

Data lights the way

The same run, two pictures. A few samples miss what happens in between. Continuous data catches it as it happens.

Conventional a few samples expected drop missed, run drifted Cellium continuous data ⚠ Drop detected in time one run → Illustrative
Work with us

Take part in the biotech revolution.

We work with a small group of partners on pilot projects and joint development.

The Cellium systemOpen for pilot partners
Pilot projects

Tell us about your project

Share your challenge and goals. Together we find the best way Cellium technology can move your project forward.

Partnerships

Join us

Research groups and companies that want to shape the next generation of bioprocess equipment.

  • Joint process development
  • Research and industry collaborations
About us

Bioprocess engineers building the lab of the future.

Cellium Biotech started at the Technical University of Munich (TUM), where we saw how much time bioprocess teams lose to slow, data-poor experiments. From Straubing, we build the bioreactor we always wanted: powerful AI, simple to use.

Pablo Gómez RossiCEO & Founder

Bioengineer with a Master's in Chemical Biotechnology from TUM. Leads product development: hardware, electronics, software and the digital twin.

Merve GezenCo-founder · Bioprocess & sustainability

Bioengineer with a Master's in Chemical Biotechnology from TUM. Expert in bioprocess development, techno-economic analysis and life-cycle assessment.

Contact

Let's take your process to scale.

Germany